1. Field of the Invention
The present invention relates to a technique of correcting an image shift caused by camera shake in an apparatus which records or plays back a video signal.
2. Description of the Related Art
Video recording apparatuses, such as a camera-integrated video recorder (to be referred to as a video camera hereinafter), which record a video signal obtained by sensing an object suffer a “camera shake phenomenon”. That is, a sensed image shifts vertically and horizontally owing to small vibrations of the hand and arm. If the user keeps seeing an image suffering camera shake, he feels sick as if he got seasick. To prevent this, recent video cameras have a function of detecting camera shake to suppress an image shift caused by camera shake.
As a conventional method of detecting camera shake, there are known a method of obtaining a motion amount and motion direction from a previous frame by image processing, and a method of detecting vibrations of a video camera using a gyroscope. As a method of suppressing the “camera shake phenomenon”, there are known a method of controlling the read area of image data rasterized in a memory, and a method of controlling the lens orientation in accordance with vibrations. There is also known a method of controlling the effective area of a video signal obtained by an image sensor such as a CCD (Charge Coupled Device) image sensor or a CMOS (Complementary Metal Oxide Semiconductor) image sensor.
However, according to the conventional camera shake suppressing method, even when panning is done to capture an image while horizontally moving a video camera, or tiling is done to capture an image while vertically moving it, the camera shake suppression acts, and the user cannot obtain an image he wants. According to the method of controlling the read area of image data rasterized in a memory or controlling the effective area of a video signal obtained by an image sensor, the effective area becomes narrow to suppress an image shift caused by camera shake. Data of pixels which become ineffective are compensated by image processing such as resolution conversion, degrading the image quality.
To solve these problems, Japanese Patent Laid-Open Nos. 6-189176 and 10-42233 propose a technique capable of selecting whether to validate or invalidate camera shake correction processing in accordance with the shooting status. According to this technique, a vibrating state detected in a video recording apparatus is recorded together with a video signal, and camera shake correction processing is executed in playback. Conventional camera shake correction processing is not adopted in all video cameras. Thus, Japanese Patent Laid-Open No. 4-268885 proposes a method of performing camera shake correction processing by a playback apparatus, and performing scan line interpolation to correct an image quality degraded by image processing such as resolution conversion, thereby obtaining a high-quality image signal.
The above-mentioned camera shake correction processing by a playback apparatus suppresses camera shake by controlling the effective area of image data rasterized in a memory in accordance with the motion amount obtained in real time. For this reason, the display size and the enlargement ratio in resolution conversion processing change every frame or field, and image data cannot be played back with a stable size and image quality. To ensure a stable size and image quality, the effective area in a memory needs to be set small in advance in accordance with a corresponding motion amount, similar to the conventional camera shake suppressing method using an image sensor. In this case, image data can be played back with the same size and image quality as long as the motion amount falls within a predetermined range. However, enlargement processing is always executed with a maximum correction amount regardless of the motion amount. Further, no camera shake correction processing can be done for a motion amount exceeding an assumed range.